On 1/26/2023 8:05 AM, Steinar H. Gunderson wrote: > I'm in the Chromium repository; it can be checked out at > https://chromium.googlesource.com/chromium/src.git (you don't need the > sub-repostiories). HEAD is pointing to 4e0db738b37c. git 2.39.1. > > When I run > > git log --raw --follow base/third_party/xdg_user_dirs/xdg_user_dir_lookup.h > > this is the first commit that it lists (snipped): > > commit 5d4451ebf298d9d71f716cc0135f465cec41fcd0 > [...] > :100644 000000 9e81e1b53029f 0000000000000 D base/third_party/xdg_user_dirs/xdg_user_dir_lookup.h > > This indicates that the last thing that happened to the file is a delete. > However, the file isn't deleted; it's alive and well. git log without > --follow does not list this commit at all. > > So either git log --follow is listing a delete commit that doesn't make > sense, or it's missing whatever commit put it back into place afterwards. It's actually a third option: it was deleted but also renamed in an independent point in history, but the delete is more recent "in time" that it shows up first, and the merge that resolves the issue doesn't show up at all. You can see this when using --full-history --simplify-merges --graph, since it will explore enough of the graph to recognize that deletion while also showing how things got merged strangely: $ git log --summary --full-history --simplify-merges --graph -- base/third_party/xdg_user_dirs/xdg_user_dir_lookup.h * commit 1e78967ed2f1937b3809c19d91e7dd62d756d307 |\ Merge: 5d4451ebf298 9e9b6e8ee772 | | Author: grt@xxxxxxxxxxxx <grt@xxxxxxxxxxxx@0039d316-1c4b-4281-b951-d872f2087c98> | | Date: Tue Jul 19 16:14:55 2011 +0000 | | | | FileManagerDialogTest.SelectFileAndCancel flaky. | | | | BUG=89733 | | TBR=robertshield@xxxxxxxxxxxx | | TEST=browser_tests | | Review URL: http://codereview.chromium.org/7447001 | | | | git-svn-id: svn://svn.chromium.org/chrome/trunk/src@93027 0039d316-1c4b-4281-b951-d872f2087c98 | | * | commit 5d4451ebf298d9d71f716cc0135f465cec41fcd0 |/ Author: jbauman@xxxxxxxxxxxx <jbauman@xxxxxxxxxxxx@0039d316-1c4b-4281-b951-d872f2087c98> | Date: Tue Jul 19 15:45:49 2011 +0000 | | Roll ANGLE r704:r705 | | BUG= | TEST=try | | Review URL: http://codereview.chromium.org/7375016 | | git-svn-id: svn://svn.chromium.org/chrome/trunk/src@93026 0039d316-1c4b-4281-b951-d872f2087c98 | | delete mode 100644 base/third_party/xdg_user_dirs/xdg_user_dir_lookup.h | * commit 9e9b6e8ee77229781fa8581b7f46413024404a5f Author: thestig@xxxxxxxxxxxx <thestig@xxxxxxxxxxxx@0039d316-1c4b-4281-b951-d872f2087c98> Date: Wed Dec 2 08:45:01 2009 +0000 Move some XDG code from chrome to base, make DIR_USER_CACHE generic rather than Chromium specific, and clean up a few headers. BUG=none TEST=none Review URL: http://codereview.chromium.org/449048 git-svn-id: svn://svn.chromium.org/chrome/trunk/src@33565 0039d316-1c4b-4281-b951-d872f2087c98 create mode 100644 base/third_party/xdg_user_dirs/xdg_user_dir_lookup.h (Note: I didn't include --follow here as it filtered the --graph output in a strange way, including dropping the merge commitswhich was confusing to me.) In conclusion: while I agree that this output is confusing, it is due to the interesting shape of the commit history, and not actually a bug in --follow. Thanks, -Stolee
